CHARACTERISTICS OF THE NON-EXPLOSIVE OXIDATION OF

Role of Formaldehyde in Combustion. RALPH KLEIN and LOUIS J. SCHOEN. 1958,58-68. Abstract | PDF | PDF w/ Links. Cover Image ...
0 downloads 0 Views 1MB Size
June, 1929

NON-EXPLOSIVE OXIDATION OF PROPANE AND BUTANES

1839

adding acetone to a concentrated alcoholic solution. As thus obtained the hydrochloride melts at 219-220”. Calycanthine hydrochloride melts a t 217O, but a mixture of the two began to sinter a t 205” and melted completely at 208”. The free base, which will be designated P-alkaloid, was only obtained as a colorless oil, readily soluble in ether. Seeding with a-alkaloid did not induce crystallization. This base also gives a red color with Ehrlich’s reagent. Isolation of Xylose from the Epicarp of the Seeds.-The residue after thorough extraction consisted of the inner fleshy portion of the seed together with the still tough and dark brown epicarp in large flat pieces. By virtue of a difference in density, the two parts could be easily separated by a process of flotation with water, the epicarp sinking. When most of the fleshy portions had been removed by this means, the residue of hulls was hydrolyzed with 2% boiling sulfuric acid for three hours, neutralized with calcium carbonate, filtered, the solution evaporated t o a thick sirup in ‘uucuo and the latter extracted once with a large volume of absolute methanol. The extract, on evaporation, yielded about 6% of almost colorless crystals which, after two recrystallizations, melted a t 150’ alone or admixed with a genuine specimen of I-xylose. It may here be mentioned that the epicarp from C. Aoridus under similar treatment yielded the same sugar in approximately the same amount.

Summary 1. The alkaloid, calycanthine, has been isolated from Meratia praecox,

Rehd. and M’ils., a shrub native to Asia. Its identity with the alkaloid from Calycanthus floridus, L., has been established. 2 . Two other alkaloids were isolated in small amounts, but the investigation of these is delayed until more material is available. 3. The existence of glucose, or a sugar which gives glycosazone, in the seeds has been shown. 4. Xylose was isolated from the epicarp of the seeds by hydrolysis with dilute sulfuric acid. NEW HAVEN,CONNECTICUT [CONTRIBUTION FROM THE LABORATORY OF PHYSICAL CHEMISTRY, UNIVERSITY ]

PRINCETON

CHARACTERISTICS OF THE NON-EXPLOSIVE OXIDATION OF PROPANE AND THE BUTANESL BY ROBERT N. PEASE~ RECEIVED JANUARY21, 1929

PUBLISHED JUNE 6 , 1929

The reactions between hydrocarbon vapors and oxygen are of interest for a number of different reasons. Information as to their nature is of This paper contains results obtained in an investigation on “Catalytic Methods Applied to Petroleum Hydrocarbons” listed as Project No. 7 of American Petroleum Institute Research. Financial assistance in this work has been received from a research fund donated by Mr. John D. Rockefeller. This fund is being administered by the Institute with the coiiperation of the Central Petroleum Committee of the National Research Council. Professor Hugh S . Taylor of Princeton University is Director of Project No. 7 . American Petroleum Institute Research Associate, Princeton Uaiversity.

1840

ROBERT N. PEASE

Vol. 51

importance in connection with the utilization of hydrocarbons, especially in connection with their combustion; and from the purely chemical standpoint the mechanism of these reactions demands investigation. Other investigators have studied the oxidation of methane, ethane, ethylene, acetylene and some of the higher paraffins. We have chosen to investigate the oxidation of the intermediate paraffins, propane and the butanes. The investigation has been comprehensive rather than exhaustive. I n particular, the nature of the liquid products has been disregarded and conclusions are based entirely on the analyses of gaseous reactants and products.

Apparatus and Procedure The flow method has been used. Known volumes of reactants a t known rates were passed into a reaction tube and the corresponding off-gas collected and analyzed. The gases passed from cylinders through calibrated flow meters (equipped with overflows to maintain constant head) and calcium chloride drying tubes, into ‘/4-inch pyrex tubes which extended into the electric furnace used for heating. These tubes were joined within the furnace to a short length of 2-mm. capillary which served as a mixing chamber, and this in turn was joined to the reaction tube. The gases then issued through a length of 2-mm. capillary to which was sealed a three-way stopcock, by means of which communication could be made either with the air or with the collecting system. The latter consisted of a trap of 5-cc. capacity and a 500-cc. mercury aspirator bottle. This was a pyrex flask to which had been sealed a three-way stopcock, The whole of the off-gas was collected in this bottle during a measured time interval, the rate of inflow of reactant gases being controlled by careful regulation of the rate of outflow of mercury from the bottle. During collection of the off-gas, the trap was cooled in carbon dioxide-ether mixture to remove liquid products. Condensable gases were, of course, bapped simultaneously. These were subsequently evaporated by allowing the trap to warm up to room temperature. The gases were then mixed by expansion and compression and submitted to analysis. After analysis the total volume of off-gas was measured and the volumes of the individuals were calculated. All gas volumes were reduced t o 25” and 760 mm. Analysis was carried out with a tVilliams apparatus to which had been added a copper oxide combustion tube. The confining fluid was water. Carbon dioxide was determined by absorption in potassium hydroxide solution, unsaturated hydrocarbons in dilute bromine water (less than 0.2% of bromine) and oxygen in alkaline pyrogallol. Carbon monoxide and hydrogen were obtained by oxidation with copper oxide a t about 300” followed by absorption of carbon dioxide. The saturated hydrocarbons were burned over copper oxide a t about 550”,with subsequent absorption of carbon dioxide. A chromel-wire-wound furnace jacketed with magnesia pipe covering was used. Its interior dimensions were 30 inches length and 4 inches diameter. The wire was wound on a piece of stovepipe covered with asbestos paper. The ends of the furnace were packed with loose magnesia, the central third constituting the actual heating space. Temperatures were measured with a single-junction chromel-alumel thermocouple, the hot junction being a t the center of the furnace. The gases were obtained from cylinders. Their compositions were as follows Oxygen-02, 98 5%; “CHI,” 1.0%; Nz, 0.5% Nitrogen-N2, 99.0%; “CHa,” 0.5%; 0 2 , 0.5% Carbon dioxide-CO2, 100.0% Hydrogen-Hz, 98.5%; “CH4,” 1.0%; 0 2 , 0.5%

June, 1929

N O N - E X P L O S I V E OXIDATION

OF P R O P A h X AND BUTANES

1841

Methane-CH4, 95.5%; Nz, 4.5% Propane-GHa, 98.5%; “CHI,” 1.0%; Nz, 0.5% n-Butane-C~H~~, 99.0%; [“CH4,” (-1.0%) 1; Nz, 2.0% Isobutane-C4Hlo, 95.5%; “CHI,” 2.0%.; 02, 1.0%; Nz, 1.0% The boiling points of the normal and isobutane were nearly constant and close to -1 and - 10 O, respectively, thus further ensuring their purity. These and the other hydrocarbons were originally obtained from natural gas. The experimental procedure was as follows The furnace was brought to temperature and the gases were allowed to flow through the system for one to two hours to ensure steady-state conditions. The off-gas was then sent through the trap and connecting tubes for one minute The carbon dioxide-ether mixture was applied to the trap (which was shut off at both ends), and one minute later connection was made through the trap with the gas collector. The whole of the off-gas was then collected for a measured time interval. At the end of this time the trap was warmed to room temperature and the condensed gases were added t o the main body, which was mixed and analyzed. The flow rates and sampling times were so adjusted that in all experiments 400 cc. of gas was introduced while the sample was being collected.

Nature of Experimental Data and Interpretation Analytical Data.-The experimental results have been expressed as follows. The gas absorbed in caustic has been called carbon dioxide. It undoubtedly consists in part of formaldehyde and probably volatile alcohols and acids. Gas absorbed by bromine water has been called simply unsaturated compounds. These are doubtless chiefly olefins. The oxygen, carbon monoxide and hydrogen determinations are unequivocal. The data on saturated hydrocarbons have been expressed in terms of the principal hydrocarbon and methane. This is, of course, arbitrary. Fortunately, however, analyses of the off-gas from the majority of the experiments indicated that the hydrocarbon introduced was the chief one present-no large quantity of new hydrocarbon had resulted from the oxidation. The same holds for the cylinder gases. We have then to compare the volumes of carbon dioxide, unsaturates, oxygen, carbon monoxide, hydrogen, methane and original hydrocarbon in the off-gas with the volumes of oxygen and hydrocarbon introduced. We find that in general the principal gaseous products of reaction are carbon monoxide and unsaturates, sometimes one and sometimes the other predominating. Methane appears in increasing quantities as the furnace temperature is raised. Hydrogen and carbon dioxide, however, are only rarely important products. The appearance of unsaturates, methane and hydrogen, all in amounts increasing with the furnace temperature, recalls the dissociation reactions of hydrocarbons, for example CsH8 +CzH4 C3Hs +C3He

+ CHI + Hz

In most experiments, however, furnace temperatures were far below those required to produce appreciable dissociation by themselves; but off -

1842

R O B E R T N. P E A S E

Vol. 51

setting this there is the exothermicity of the oxidation reactions. (The question of temperature is discussed below.) We choose therefore t o regard the hydrogen and methane as products of dissociation of the hydrocarbon and to associate with them equivalent quantities of unsaturates. We therefore account for a part of the hydrocarbon reacting in terms of thermal dissociation, designating this as “Reaction I. Unsaturates by Dissociation.” The amount of this type of reaction is given by the sum of the volumes of methane and hydrogen. Thermal dissociation as thus evaluated only accounts for a portion of the unsaturates. The remainder must obviously be a product of oxidation. We cannot say what reactions are involved here, but we suggest that they be regarded as oxidations of those portions of the molecules which would be split off in dissociation leaving olefins. Thus, for example

+ HZ +C4Hs + HzO + CHd +CaHs + CHsOH +C3Hs + HCHO, etc. C4Hia +CzHa + CzHfi +CzH4 + C2HsOH +C Z H + ~ CHXHO. etc.

C4Hio +CiHs C4Hio +C3Ha

These then are products of incipient dissociation completed by oxidation of the more saturated product of dissociation. We designate these reactions as of the type “Reaction 11. Unsaturates by Oxidation.” The amount of this type of reaction is given by the total of unsaturates less the sum of the volumes of methane and hydrogen. Finally there remains that portion of the reacting hydrocarbon over and above the equivalent of unsaturates produced. I n accounting for this we note that in those cases (low temperatures, empty reaction tube) in which little unsaturates, methane and hydrogen are formed, carbon monoxide is produced in amount roughly corresponding t o the volume of hydrocarbon consumed, two volumes of oxygen disappearing simultaneously. This appears to be indicative of the type of reaction which Edgar and Pope3 have found in the case of the higher hydrocarbons and of which an example would be

+ +GHsCHO + HzO + Oz +CO + HzO + CH3CH0, etc.

CaHs C2H6CH0

0 2

I n any event we take care of the hydrocarbon unaccounted for in Reactions I and I1 by a third type, an oxidation which is designated “Reaction 111. Aldehyde Reaction.” The amount of this is obtained by subtracting the volume of unsaturates produced from the volume of hydrocarbon consumed. We analyze our results then in terms of three reaction types of which the following would be examples: Type I, Unsaturates by Dissociation, e. g., C3Hs -+ C3H6 Hz; Type 11, Unsaturates by Oxidation, e. g.,

+

* Edgar and Pope, Swampscott Meeting, 1928.

June, 1929

+ +

NON-EXPLOSIVE OXIDATION OF PROPANE AND BUTANES

+

1843

C3Hs ' / 2 0 2 + C3He H20; Type 111, Aldehyde Reaction, e. g., C3Hs 202 -+ CO(+ CH3CH0 2H20). This establishes a basis for the comparisons of the effects of altering the variables. We do not insist that the possibilities are exhausted by the above types or that the method of calculating the relative amounts is infallible. Methane and hydrogen may well be formed in part by dissociation of intermediate products; the unsaturates may not represent olefins exclusively. Especially, the arbitrary method of representing the hydrocarbons as consisting only of the original and methane doubtless introduces errors. We have, however, a definite basis for the presentation and discussion of the data and we are inclined to think that the scheme and its application are not wide of the mark.

+

Temperature.-With respect t o the reported temperatures, the following is to be kept in mind. These reactions are as a whole exothermic and take place rapidly. Exposure times are of the order of a few seconds. The gases are, therefore, unquestionably a t a higher temperature than the surroundings. However, the whole question of temperature under such conditions is decidedly vague. Newly-formed molecules are necessarily highly energized. No matter what the total reaction rate may be, the unchanged molecules with which these product molecules iirst come in contact are subjected to far more extreme conditions than would correspond to many hundreds of degrees general rise in temperature. Only under such conditions that every newlyformed molecule made many wall collisions before meeting an unchanged molecule would anything corresponding to an isothermal state be set up. In our preliminary work much time was spent in attempting to adjust temperatures down the reaction tube by using a number of independent heating elements. Arrangements were made for determining temperatures a t various positions. It was always found, however, that if reaction took place a t all, the thermocouple a t the entrance to the reaction tube registered from 25 to 50" above that somewhat farther down. We came, therefore, to the conclusion that a better arrangement was simply to place the reaction tube in a constanttemperature bath-the furnace-over its entire length, and to measure and record its temperature. This is a t least definite and should be reproducible. The reaction then takes place in a manner that is partly adiabatic and partly isothermal. Since the gases were heated prior to mixing, the recorded temperatures are also the initial temperatures of the reaction mixture. Heating Time.-Some uncertainty exists as t o the exact times of heating in these experiments. When there is no change in the number of molecules on reaction, the temperature is known and constant and the flow is uniform across the reaction tube, the heating time is obtained by dividing the volume of the tube by the flow rate corrected to the temperature and pressure a t which reaction is occurring. However, in these reactions, the number of molecules changes on reaction in so far as dissociation is a factor and the actual temperature of the reacting mixture is uncertain. Instead of trying to estimate these effects in separate experiments, and adjusting the flow rate t o give a uniform contact time, we have been content to maintain a constant flow rate (measured a t room temperature and pressure) throughout any one series of experiments. The average heating time in any series has then been taken as the ideal heating time a t a temperature of 472" (745" A., which is 21/2 times the average room temperature, 298"). As experiments were conducted over the range 300-625", this temperature is an average for all experiments. Rooni temperature might as well have been the basis of cdculation, but the above gives more nearly the actual heating times.

1844

ROBERT N. PEASE

Vol. 51

The larger empty reaction tube had a volume of 21.6 cc. and the total flow rate at 25" and 760 mm. was 40.0 cc./min. The average heating time was therefore

21'6 X 60 = 13.0seconds 743 40.0 X 2 ~ 8 The packed tube had a volume of 15.8 cc., and the total flow rate was 30.0 cc./min. The average heating time was 12.6 seconds. The results of experiments in the two large tubes are therefore directly comparable as to heating time. The smaller, empty tube had a volume of 2.6 cc. and the flow rate was 40.0 cc./min. The average heating time was 1.6 seconds. Heating times in this tube are therefore approximately one-eighth as long as in the larger tubes. The large, empty tube was 1.65 em. in internal diameter, 11 cm. long with 2-mm. walls. The small empty tube was 0.5 em. in internal diameter, 10 cm. long with 1-mm. wa!ls. The large, packed tube was 2 cm. in internal diameter, 11 cm. long with 2-mm. walls. It contained 46 g. of broken glass of 6-20 mesh. The glass surface in this tube was 10 to 15 times that in the empty tube.

Results and Discussion Wall Action.-The first point t o be settled was, of course, the influence of the walls of the reaction tube on the reaction. The result was unexpected. In order to determine whether or not the wall material (pyrex glass) catalyzed the reaction, a comparison was made between reactions taking place in an empty tube and in one packed with broken pyrex glass. (The tubes were cleaned with hot concentrated nitric acid and rinsed with water before use.) The result of the comparison was that whereas in the empty tube reaction between oxygen and propane was complete a t 375", in the packed tube there was no reaction below 500". The glass packing therefore strongly suppresses reaction rather than accelerates it, and we must conclude that the walls have a predominantly negative rather than positive catalytic effect. This is in harmony with the observations of Hinshelwood and Thompson on the hydrogen-oxygen c ~ m b i n a t i o n . ~ Related observations are those of Gautier and H61ier5on the explosibility of hydrogen-oxygen mixtures in a tube packed with porous porcelain. This inhibitory influence of wall surface undoubtedly has to do with the more efficient transfer of energy from the reaction mixture to the surroundings. On the older view the initial temperature is more nearly maintained in presence of the glass packing; on the newer view excited molecules of product are de-activated by the walls instead of transferring excitation to fresh molecules of reactant. We will return t o this point later. There was still the possibility that in addition to this inhibitory influence of wall, which was regarded as physical in nature, the wall material was capable of exerting a positive catalytic effect on the oxidation. If this 4 Hinshelwood and Thompson, Proc. Roy. SOC.(Lolzdon) 118A, 170 (1928); see also Pease and Chesebro, Proc. Nut. Acad. Sci., 14,472 (1928). 6 Gautier and Hltlier, Compt. rend., 122,566 (1896).

June, 1929

N O N - E X P L O S I V E OXIDATION

OP PROPANE: AND BUTANES

1845

were the case i t should be possible to poison the active surface and obtain a further reduction in the amount of reaction. Poisoning the surface amounts to covering it with some material which is itself inactive. Alkali halides suggested themselves as substances which might fulfil this object. Accordingly the tubes were rinsed out with 20y0 potassium chloride solution, drained and again put to use. It was found that there was no alteration in the amount of reaction of oxygen and propane in the empty tube. In the packed, “coated” tube, however, the temperature a t which reaction began was raised from around 500 to 575’) and reaction was still incomplete a t 623”) whereas prior to coating it had been nearly complete a t 5.50”. We have then to deal not only with an inhibitory action of the walls but also with a positive catalytic effect which, however, can be markedly decreased (and possibly eliminated) by poisoning. In the large, empty tube, on the other hand, we are safe in assuming that the observed reaction is almost exclusively homogeneous. We will designate the reaction taking place in the large, empty, coated tube as the “free reaction,” and that in the large, packed, coated tube as the “inhibited reaction.” It is of interest to consider these two in some detail. Because i t is in some respects the simpler and more normal of the two, the inhibited reaction will be first discussed. The Inhibited Reaction Propane.-Data on the reaction between propane and oxygen in the packed, coated tube are given in Table I. (In all these experiments close to 400 cc. of gas was introduced while the sample of off-gas was being collected, the actual amount being, of course, used in the calculations. For the sake of clearness, the amounts in round numbers are entered in TABLE I

Temp., OC.

575 600 625 575 600 625 600 625

REACTION O F PROPANE AND OXYGEN I N PACKED, COATED TUBE % ’ CiHa introduced to reaction, I I1 111 unsats. unsats aldeGas in, cc. Reacting, cc. ---Formed, cc.---hyde by by 0 2 C3H8 Oz CaHe Unsats. CHI Ha CO COz dis. ox. react. Total

300

100

100

300

Nz 100

8 33 35 6 39 84

300

13 27 32 27 78 144 33 45

4 3 20 1 25 14 14 10 64 33 119 105 22 40

2 1 1 4 5 2 3 2 2 1 1 1 3 3 2 1 3 5

32 3 32 17

5 5 17 4 12 35

0 15 8 1 9

4 5

13 27 32 9 26

4

8

47

8 7 7

11 15

11 15

Before coating with KCI @I

550

300 100

100 300

287 95

88 146

15 101

8 13 167 49 58 11 41 6

15 24

0 11

74 15

89 50

1846

ROBGRT N. PGASE:

Vol. 51

the tables under “Gas in.” In addition, the temperature, volumes in cc. of hydrocarbon and oxygen reacting and of gaseous products formed and the percentages of total hydrocarbon introduced which have reacted according to the reaction types in which the results have been expressed are given.) The first point to be noted is that approximately the same fraction of propane reacts whether the initial propane-oxygen ratio is 3 t o 1 or 1 to 3. The conversion of propane is therefore approximately first order with respect to its own concentration. Since a large proportion of what reacts appears as dissociation products, one might suspect that the chief reaction was a purely thermal dissociation. When, however, nitrogen is substituted for oxygen as diluent, it is found that only one-third to one-half as much propane reacts. Hence it is clear that the dissociation has been for the most part excited by the simultaneous oxidation reactions. It is nevertheless significant that one has to approach conditions conducive to thermal dissociation before any reaction occurs. The oxidation reaction is evidently contingent on the possibility of dissociation, and a t the same time promotes dissociation. It is as though the underlying reaction were still the unimolecular dissociation of the hydrocarbon but that the degree of activation requisite for reaction were lowered. It is t o be noted that little if any carbon monoxide, carbon dioxide or hydrogen is formed. Since less hydrogen is obtained from the oxidation than from the dissociation, it is evident that some has been burnt-under circumstances, i t may be added, such that little hydrogen alone would have reacted. There is fairly clear indication here of removal of hydrogen directly from the hydrocarbon molecule by oxidation. This was given as an example of one of the type reactions, and for this reason. The absence of carbon monoxide and carbon dioxide shows that such carbon-oxygen linkages as have been formed have issued in liquid products. The marked difference in this respect between experiments before and after coating with potassium chloride will be noted. It is a peculiarity, attributable to the unimolecular character of the reaction, that more oxygen is consumed from the mixture which is deficient in oxygen. It should be noted also that the mixtures richer in oxygen, although the more explosible because of their greater potential energy content, are under these conditions the less reactive. The Free Reaction Propane.-We turn now t o a consideration of the very different type of reaction between propane and oxygen which takes place first a t 300-400° (as compared to 575-600’) in our larger, empty tube. Carbon monoxide is now the principal gaseous product. The reaction is of a highly spontaneous character and gives every evidence of being of the chain type. Results are presented in Table 11.

fune, 1929

NON-EXPLOSIVF$ OXIDATION OB PROPANE AND BUTANES

1847

TABLE I1 REACTIONS OF PROPANE AND OXYGENIN THb LARGER, EMPTY, COATED TUBE % CaHs introduced to reaction, Temp., OC.

Gas in, cc. 0% CaHs

300 325 350 375 400 425 450 300 325 350 375 400 425 300 325 350 375 400 450 500 550 600

300

375

300 200 100

200

100

Reacted, cc. 0 2 CsHs

----Formed. CC.--Unsats. CH4 Hz CO COa

I I1 I11 unsats. unsats. aldeby hyde dis. react. Total

%.

100

200

300

100 200 300

Nil 119 51 115 50 97 44 67 42 Explosive XU 183 98 192 100 193 105 187 107 Explosive Nil 93 66 96 70 94 74 94 81 94 101 93 137 96 151 94 178 124 189 93

Nil 10 17 16 19

0" 9 59 14 0 4 58 11 0 9 50 9 0 4 35 5

9 4 9 4

1 4 3 5 3 14 34 52 7 29 45 15 24 43

Nil 20 24 31 37

3 4 91 0 3 90 0 11 95 4 5 92

4 2 6 5

7 11 11 15

41 40 38 36

52 53 55 56

2

18 16 17 15 13 16 15 16

23 24 27 28 34 48 53 62

35 36 14

50 51 26

Nil 15 24 27 37 63 91 109 131

15 17 14 11

6 5 4 5 5 5 4 3

5 9 21 25 40

3 4 5 8 12 11 13 6

Before coating with KC1 49 15 0 6 56 11 99 30 1 11 88 13 73 32 6 7 39 4

6 6 5

9 9 7

2 5 8 8 15 48 53 81

5 6 5 6 12 12 19 33

45 45 40 40 40 48 51 54

4 5

In these and many other experiments with excess of oxygen, the saturated hydrocarbon in the off -gas averaged slightly higher in carbon than the original hydrocarbon. There would thus be no indication of the presence of methane. Small quantities of saturated hydrocarbons of higher molecular weight than the original appear to be formed under these conditions. The discrepancy never exceeded a few cc. and no attempt has been made to allow for it.

The reaction does not follow any conventional scheme. It starts abruptly within a 25' temperature interval, and, in the case of the 3C3Hs: 1 0 2 and 2C3Hs: 202 mixtures, runs immediately to complete consumption of the oxygen. Somewhere between C3Hs:02ratios of 2:2 and 1:3 there is another abrupt change such that a t the lower propane concentrations only one-half to one-fourth of the oxygen reacts, although the start of reaction as the temperature is raised is still abrupt. Moreover, the reaction has under such conditions a quite definitely negative temperature coefficient, although it eventually becomes mildly explosive. At first sight the diminished reactivity of mixtures low in propane ap-

1848

VOl. 51

ROBERT N. PEASE

pears to be in harmony with the results of the corresponding experiments in the packed tube a t higher temperatures. There is, however, more than a suspicion that in the empty tube excess of oxygen rather than deficiency of propane is the important factor in limiting the reaction. Experiments were carried out in which one-third of the oxygen of the 1C3Hg:302 mixtures was replaced by some other gas giving mixtures of the composition 1C3Hs:202:1 diluent. The diluent gases were nitrogen, carbon dioxide, methane and hydrogen. Results are presented in Table 111. EXPERI?LENTS O X RIIXTURES Gas in, Temp OC.

375

I

Diluent

02

CH4

Nz COz H?

425

Oz CHi N2

cox

H2

TABLE 111 COMPOSITION 1CsH,3:202:1 DILUENTIN THE LARGER, EMPTY, COATED TUBE 100 c c . ; 02, 200 c c . ; diluent, 100 cc. 5% CaHsintroduced to reaction,

OF

I I1 unsats. unsatu.

ox.

I11 aldehyde react.

115 109 109 94 93

50 53 47 46 42

ii 15 1 4 13 12

o

4 58 11 0 3 47 8 0 6 4 5 9 0 4 45 (15) 0 (2) 47 8

4 3 6 4 (2)

i3 12 8 * 9 (10)

34 39 35 35 31

52 54 49 48 43

67 31 11 12 Nil

42 19 12 8

19 13 10 6

0 0 0 0

4 3 1 4

15 10 9 2

24 6 2 2

43 19 12 8

Reacted, cc. 0 2 C3Ha

Formed,cc---Unsats CHI Hz CO COa

C---

4 35 5 3 12 1 1 8 1 4 7 (11)

by

dis.

by

Total

The substitution of diluent for part of the oxygen makes very little difference in the amount of reaction a t 375". There is a small decrease in the order 0 2 , CH4, X2, C02, Hz. At 425", however, this effect becomes so pronounced that, in the presence of hydrogen, reaction is almost entirely suppressed. These diluent gases clearly have specific inhibitory effects on the oxidation, the effect being more pronounced a t the higher temperature. The similarity of the results when oxygen is the diluent leaves little room for doubt that this substance itself is as well inhibitor as reactant, and especially that the negative temperature coefficient observed with excess of oxygen is intimately related to the inhibitory action of this substance. That propane does not have a similar effect may be seen by comparing the results of experiments with 2CBH8:202 mixtures (Table 11) with the above. So far as is known, no similar observations on homogeneous gas reactions have been noted except those involved in anti-knock action, which refer, of course, t o precisely this type of reaction-the oxidation of hydrocarbons. The characteristics of the free reaction convince us that we are dealing with a chain mechanism. The reactions in question are undoubtedly

June, 1929

NON-EXPLOSIVE OXIDATION OF P R O P A N E AND B U T A N E S

1849

those suggested by Edgar and Pope3 as the reactions of oxidation of the higher paraffins, and may be represented by the equations

+ CzHsCHO f H20 + O2+ CH3CH0 + HeO + CO, etc.

C3Hs CzHaCHO

0 2 +

In the case of propane the two reactions written above appear to be chiefly concerned, since in general we obtain 1 mole of carbon monoxide for 1 mole of propane and 2 moles of oxygen reacting. This is the sequence of reactions which is suppressed by diluents and KC1-coated, glass packing, and which has the chain mechanism. These reactions induce those yielding unsaturates and are the more effective in this respect as the temperature is raised. Further information as to the nature of the free reaction is given by experiments in which 50% propane-oxygen mixtures were diluted with nitrogen. Results are presented in Table IV. TABLEIV REACTION OF 50y0 PROPANE-OXYGEN MIXTURES DILUTEDWITH NITROGENIN THE LARGER, EMPTY,COATED TUBE % CsHs introduced, t o reaction, Temp., "C.

350 375 450 500 550 600 300 350 400 450 475 300 350 400 350

400

Gas in cc. Reacted, cc. ---Formed, cc.---N2 02 C3Hs Oz C3Ha Unsats. CHI H2 CO COz

300

50

50

1

by

by ox.

111 aldehyde react.

Total

0 10

23 37 28

56 82 79

dis.

Nil

J

28 27 46 40 48 37 200 100 100 Nil 73 39 Nil Nil Slightly ex76 62 plosive 100 150 150 Nil 138 69 75 53 300 50 50 Nil 200 100 100 73 39 100 150 150 138 69 0 200 200 192 100 200 300 100 50 100 5o 100 150 150 75 53 0 200 200 187 107

1

I I1 unsats. unsats.

16 0 0 12 1 22 1 4 28 3 24 12 3 35 3

32

33 35 19

9

0 433 6

4

5

32

41

37

14 17 50 6

33

6

26

65

16 21

0 1 69 11 0 3 31 4

1 2

10 13

37 23

48 38

9 16 24

0 433 6 0 1 69 11 0 3 90 17

4 1 2

5 10 11

32 37 40

41 48 53

21 37

0 3 31 4 4 5 92 11

2 5

13 15

23 36

38 56

Results with mixtures containing 50% nitrogen are particularly striking. Considerable reaction took place a t 350°, but not a t 300, 400 or 450". At 475" the mixture became very mildly explosive but a sample could

1850

Vol. 51

ROBGRT N. PGASG

still be taken. The reaction rate first increases with temperature (300350'), then decreases t o zero (350-450') and finally again increases (450475"). When the nitrogen content is increased to 75% the whole of the low temperature reaction is wiped out, and only the higher temperature region (500-600 '), with its positive temperature coefficient but without the explosive character, remains. It is to be noted that when reaction does set in, the formation of unsaturates is considerable as compared to that of carbon monoxide. This is equally true for the 50% mixture a t 475'. We have seen that the reactions of propane and oxygen can be suppressed by dilution or by the use of a KC1-coated glass packing. It was desired to determine the effect of shortening the heating time. To accomplish this i t was convenient to keep the flow rate the same and use a smaller reaction tube. Accordingly a tube of the same length as the larger, empty tube (about 11 cm.) but of about one-third the diameter (0.5 cm. as against 1.6 cm.) was made up. Results are given in Table V. TABLE V REACTION OF

PROPANE AND OXYGEN IN THE SMALLER,

EMPTY,COATED TUBE yo CaH8 introduced to reaction,

Tzmp C.

600 625 500 550 600

,

Gas in, cc. On CaHs

300 100 Explosive 100 300

Reacted, cc. On CaHs

12

10

---Formed, cc.--7 Unsats. CHI Hn CO CO1

7

1

7

4

0

38 19 0 1 3 125 58 17 36 6

6 26

6 17

3

11

0

4

I I1 111 unsats. unsats. aldeb . ~ by hyde dls. ox. react. Total

1

1

1

5 52

Nil 25 93

48 150

9

Reaction is again forced into the higher temperature region with the formation of unsaturates predominating. The heating time in this smaller tube was one-eighth of that in the larger tube. This would be expected to decrease the amount of reaction correspondingly; but the observed decreases are so great as to warrant the conclusion that the decrease in tube diameter is effective of itself. Reaction only sets in a t 500-600" instead of a t 300-400°. I n the smaller tube contact between gas molecules and the tube walls is facilitated. Apparently the difference in this respect between the smaller and larger tubes is sufficient t o give the observed marked differences in reaction characteristics. There is finally one more way in which the propane-oxygen reactions can be affected. This is by reducing the pressure. Experiments were carried out in which hydrocarbon-oxygen mixtures were caused to flow through a tube of large volume a t 5 to 10 mm. pressure. The quantitative results were not very satisfactory owing to irregularity. Qualitatively, however, there was no question but that the suppression of reaction was very great. No reaction whatever took place below 600". At higher

June, 1929

NON-EXPLOSIVE OXIDATION

oE'

PROPANE: AND BUTANES

1851

temperatures there was certainly a parallelism between the amounts of hydrocarbon dissociating in nitrogen and the amount reacting in oxygen. It was not possible to conclude whether the same amount of hydrocarbon was involved in the two cases, or whether the two reactions were independent of each other. Once more, however, we find reaction finally setting in under conditions which are favorable to dissociation alone. What has been said of the low pressure oxidation of propane applies as well to that of the butanes and also of ethane. In the oxidation of propane we have then to deal with three types of reaction-that yielding unsaturates, and methane or hydrogen by dissociation, that yielding unsaturates and unidentified liquid products by oxidation and that yielding carbon monoxide and unidentified liquid products (again by oxidation). The first two types always appear together, sometimes one and sometimes the other predominating, and increase in amount as the temperature is raised. Unless the carbon monoxide reaction is taking place (as in the larger, empty tube) the reactions yielding unsaturates do not put in an appearance until the temperature reaches 550-600'. They give evidence of being more or less normal homogeneous gas reactions, first-order with respect to propane, which are subject t o excitation by the carbon monoxide reaction in amount which is small a t the lower temperatures but increases as the temperature is raised. In the carbon monoxide reaction we meet with a wholly different state of affairs. This reaction is promoted by excess of hydrocarbon but depressed by excess of oxygen. Dilution with other gases than oxygen likewise inhibits the reaction; the extent of inhibition is a specific property of the diluent, and suppression by any of the diluents (including oxygen) is greater a t higher temperatures so that the reaction under these conditions has a negative temperature coefficient. This reaction is likewise suppressed by a glass packing, particularly when this is coated with potassium chloride; it is suppressed in reaction tubes of small diameter; and i t is suppressed when the pressure is lowered to a few millimeters. These are characteristics to be expected of a chain reaction, and we do not hesitate to state that a chain mechanism is here involved.' We have, however, no evidence bearing on the nature of the chain, though the formation of aldehydes and the possibilities of peroxide formation are t o be borne in mind.' That the reactions are as a whole exothermic is shown by the fact that a jacketed thermocouple will register a temperature 25-50' higher in a small region close to the inlet end than in the rest of the tube. A spontaneous development is not surprising. Rather, what requires explanation under these circumstances is the fact that other possible reactions E

We have recently found that this reaction is also faintly luminous. See Mardles, J . Chem. Sac., 133,872 (1928)

1S52

VOl. 51

ROBERT N. PEASE

are not more influenced, in particular, the fact that the excess of hydrocarbon which is always present is not more radically and more completely decomposed. The evidence is all in favor of the conclusion that the reaction energy is not as a whole available for heating the gases but is largely retained in some special form by product molecules to be given up to the walls or to be devoted by a specific mechanism to the continuance of the particular reaction t o which i t is due. Reactions of the Butanes with Oxygen.-It remains now t o present data on the reactions of the butanes with oxygen under conditions corresponding to those of the propane-oxygen reactions which have aready been presented and discussed in detail. The relations are essentially the same as with propane. Reaction is in general more ready and suppression is more difficult. Of the two, n-butane is the more reactive. The Inhibited Reaction.-Results with .the packed coated tube are presented in Table VI. TABLE VI REACTIONS OF Temp., “C.

Gas i n , cc. H.C.

Ox

450 500 550 450 500 550

300

100

100

300

400 450 500 400 450 500

300

100

100

300

OXYGEN IN TITE PACKED COATED TUBE % H. C. introduced t o reaction, I I1 I11 unsats. unsats. aldeReacted, cc. Formed, cc.---by by hyde Os H.C. Unsats. CHP Hs CO COa dis. ox. rescts. Total

TKE

BUTANESAND

----

40 14 109 42 197 69 45 25 87 73 97 107 56 140 183

15 46 63

Isobutane 11 0 4 10 22 0 3 42 31 14 11 106 28 0 6 8 63 23 3 15 89 74 2 5 11 20 31

8 18 24 11 15 12

4 3 26 2 9 27

7 3 1 20 21 6 40 8 0 1 13 3 5 6

4 44 76 0 25 38

n-Butane 4 4 7 7 0 4 46 33 8 7 80 33

8 4 15

3 4 1 15 26 16 33

5 45 64

Xi1 58 35

34 9 1 4 15 3 8 0 1 1 88 85 78 22 6 23 17 10 17 2 29 a The interpretatioa of the analyses for saturated hydrocarbons in the off-gas are the more uncertain in these cases because of the probability that ethane is also present.

A comparison of these with the corresponding results for propane shows that whereas the latter only reacted a t 600“ with very little carbon monoxide formation, the butanes are reacting a t 500 o and considerable carbon monoxide is produced. The packing, therefore, does not repress the “aldehyde reaction” so readily in these cases. A finer packing might perhaps be more effective. -4s between the two, reaction development is the more ready in %-butane but the differences are not great, particularly when oxygen is in excess.

June, 1929

N O N - E X P L O S I V E OXIDATION

OF

P R O P A N E AND B U T A N E S

1853

The Free Reaction.-Reaction between the butanes and oxygen in the larger, empty, coated tube takes place in a manner not very different from that with propane. ,4 greater proportion of oxygen is consumed and reaction passes more readily into explosion. Indeed, n-butaneoxygen mixtures of composition 2 : 2 or 1: 3 showed no region of quiet reaction. Results are presented in Table VII. TABLEVI1 REACTIONS OF THE BUTANES AND OXYGEN

Tpp., Gas in, cc. Reacted, cc. C. diluent Oz H. C. Oz H. C.

IN THE

LARGER, EMPTY,COATED TUBE

---Formed, cc.--Unsats. CHI Hz CO COz

C3Ha introducer! t o reaction, I I1 111 unsatq. unsats. aldeby by hyde dis. ox. react. Total

Isobutane 12 0 5 75 17

5

7

52

64

29 4 0 37 6 39 11 0 34 6 68 33 9 41 7 16 0 4 59 14 14 0 6 68 10

1 4 15 4 6

9 10 9 13 8

11 13 11 37 49

21 27 35 54 63

n-Butane 02 have no zone of quiet reaction Mixtures of 100 C4Hl&00 02 or 200 C~HIO-200 350 hTone 100 300 93 64 30 26 6 36 9 10 0 12 450 96 91 77 42 7 39 12 17 10 5

22 32

350 None 300 400 Explosive 350 None 100 400 450 400 N2, 100 200 H1, 100 200 425 Both diluted

100 181

62

300

100 60 98 77 100 100 100 149 51 100 160 61 mixtures explosive

With 3 : l hydrocarbon-oxygen mixtures (when all of the oxygen is consumed) the proportions of reactants and products are much the same as with propane. The 1: 3 isobutane-oxygen mixture reacts only incompletely when i t does not explode (e. g., a t 350") as in the case of the corTABLE VI11 REACTIONS OF THE BUTANES AND OXYGEN

Temp., OC.

Gas in, ce. Oz H . C.

500 550 450 500 550

100 300 Explosive 100 300

450 500 550

100

300

Reacted, cc. Oz ' H. C.

43

15

IN THE

----Formed, cc.---Unsats. CHa Ha CO Isobutane 15 0 2 14

SMALLER, EMPTY,COATED TUBE 5% C3Ha introduced t o reaction, COz

I I1 111 unsats. unsats. aldeby by hyde dis. ox. react.

Total

5

1

14

0

15

22 13 5 4 5 2 99 116

18 46 101

5 7 9

0 2 16

7 14 19

0 2 5

7 18 30

70 70 93

n-Butane 42 4 2 17 13 68 10 4 16 13 130 53 8 32 12

2 5 20

12 18 21

4 2 0

16 25 41

49 75 124

0

5 39

0 0 0 9 8 32

1854

Yol. 51

ROBERT N. PEASE

responding propane mixture, though a greater proportion of the oxygen is consumed. There is less indication of suppression by diluents, however, Data referring to the smaller, empty, coated tube are given in Table VIII. It is seen that in the smaller, empty, coated tube the butanes again react more readily than propane, and that n-butane is the more reactive of the two. In this case the system of interpretation appears to be strained somewhat by the appearance of excessive quantities of unsaturates. It is possible that the formation of ethane is responsible, or that in some cases two molecules of ethylene are formed from one of butane. A comparison between the amounts of reaction of 3 :1 butane-oxygen mixtures a t 450' in the larger and smaller empty tubes is available and is of some interest. It will be recalled that the two tubes give heating times in the ratio of 1 to 8 (1.6 and 13 sec., respectively) their diameters being in the approximate ratio of 1 to 3. Results are given in Table IX. COMPARISON O F

TABI.EIX 3:1 BUTANE-OXYGEN MIXTURESAT 450"

IN SMALLER AND LARGER,

EMPTY, COATEDTUBES Tube

Reacted, cc. 0.1 H. C.

-----Formed, Unsats. CHd

Kz

cc.---CO

Small 22 Large 100

13 100

18 68

Isobutane 0 0 0 33 9 41

70 96

49 91

42 77

n-Butane 2 17 42 7 39

Small Large

4

COz

CaHs introduced I I1 unsats. unsats. by by dts. ox.

t o reaction, % 111 alde-

hyde react.

Total

5 7

0 15

7 9

11

7 35

13 12

2 17

12 10

4 5

16 32

0

The differences are ascribed to an increase principally in the amounts of dissociation, along with a considerable increase in the amount of carbon monoxide, which, however, does not appear in the analysis of results as due to our aldehyde reaction, though doubtless this would be the correct interpretation. We conclude that there is no essential physico-chemical difference between the reactions of the butanes with oxygen and those of propane and that the same considerations are applicable. This being the case, we are inclined to believe that the same will be true of the higher hydrocarbons. Differences will necessarily appear in the study of the lower hydrocarbons, methane and ethane, because of the impossibility of certain types of reaction, and these will require special study. The appearance of carbon monoxide and water as the principal products of methane oxidation is suggestive of a similarity in some sense. In the case of ethane the slowness of the reactions when oxygen is in excess, as shown by Bone and Stockings,8 is exactly in accord with results found for propane and the 8

Bone and Stockings, J . Chew. Soc., 85,693 (1904).

June, 1929

N O N - E X P L O S W I $ OXIDATION

OF P R O P A N E AND B U T A N E S

1855

butanes. It may be added that Bone and Wheelergand Bone and Andrew10 find the same effect in the slow oxidation of ethylene and of acetylene, pointing to chain mechanisms in these cases also. These results are admittedly incomplete. What was aimed a t was a survey of the field. The points that will interest us especially in further investigations are the nature of the chain reaction and especially of the diluent effect, and the behavior of the lower hydrocarbons. It may be pointed out that the indications are that liquid products containing the same number of carbon atoms as the original hydrocarbon are not formed in quantity. The aldehyde with one less carbon atom should be obtainable in a fairly pure state by oxidation a t the lowest possible furnace temperature with excess of hydrocarbon. High temperatures and the maximum suppression give good yields of what are undoubtedly olefins, under conditions which are more convenient than those required for ordinary dissociation. I n fact, this is unquestionably a suitable method to prepare olefins from saturated hydrocarbons. We have no very clear ideas as to the nature of the reactions producing unsaturates by oxidation except that the burning off of hydrogen to give the olefin of the same chain length is one reaction, and that reactions such as

++

C~HR C3Hs

+

0 2

4

CzHi CiHa

+ CHIOH

+ HCHO + He0

are possible. We are inclined to believe that the formation of alcohols, which probably occurs, is referable to the first of these reactions and may be regarded as due t o the tendency of the hydrocarbon to undergo demethanation. The formation of acids js perhaps due t o secondary oxidation of aldehydes but we have no evidence on this point for or against. There is no question but that the proportions of these various reactions can be altered profoundly by relatively minor changes in the methods of carrying out the oxidation. This is to be borne in mind by those interested in arriving a t a more precise idea of the nature of the liquid products, or in forcing the reactions in some one particular direction. One point in particular is worth bearing in mind. Probably any catalyst will act to suppress what we have called the aldehyde reaction just as a glass packing does. On top of this effect will be superimposed the specific action of the particular catalyst employed. The role of the catalyst will thus be always a dual one. Summary 1. A study has been made by the flow method of the reactions of mixtures of oxygen with propane and the butanes. No attempt has been made t o analyze the liquid products and all conclusions are based on the results of gas analyses. Bone and Wheeler, on ethylene, J . Chm. SOC.,85, 1637 (1904). Bone and Andrew, on acetylene, ibid., 87, 1232 (1905).

1856

Vol. 51

CHARLES BUSHNELL WOOSTER

2. The reactions are classified into three types, which (with examples) are as follows: (I) Unsaturates by Dissociation, ~xample-C3H8 + CaH6 Hz; (11) Unsaturates by Oxidation, Example-C3Hs 1/202 + C3H6 H2O; (111) Aldehyde Reaction, Example-C3H8 2 0 2 ----f CO 2H20 CH3CHO. 3. Types I and I1 give evidence of being more or less normal homogeneous gas reactions which appear a t (300-600’, but are subject to induction by Type I11 a t lower temperatures in amount increasing with the temperature. 4. Type I11 may begin a t temperatures of 300-350’ and gives every indication of being a chain reaction. Its development is highly spontaneous, and a t the same time it is sensitive to relatively minor variations in the conditions of reaction. In particular, it is subject to suppression by a broken-glass packing (especially when this is coated with potassium chloride), and by use of a reaction tube of small diameter. Lowering the pressure to a few millimeters has a like effect. Dilution with oxygen, methane, nitrogen, carbon dioxide or hydrogen depresses the reaction in amount increasing in the order named; this effect is the greater a t higher temperatures. Under these conditions the reaction thus has a negative temperature coefficient, though eventually becoming explosive. Excess of hydrocarbon promotes the reaction. (3. Under suppression, there may be no reaction up to 500-600”. Types I and I1 then appear, accompanied by Type I11 in varying amount. 6. Considerations bearing on the isolation of particular reactions are set forth.

+

+

+ +

+

+

PRINCETON, r\TEV’ JERSEY [CONTRIBUTION FROM THE RESE.4RCH LABORATORY O F PHYSICAL CHEMISTRY. MASSACHUSETTS INSTITUTE OF TECHNOLOGY, No. 2221

THE STRUCTURE OF METAL KETYLS. I. THE SCHMIDLIN FORMULA BY CHARLESBUSHNELL WOOSTER~ RECEIVBD JANUARY 23, 1929

PUBLISHED JUNE 5, 1929

The highly colored products formed by the addition of one atom of an alkali metal to one molecule of certain aromatic ketones were named “metal ketyls” by Schlenk and his co-workersJ2who concluded that these substances comprised a new class of compounds containing trivalent carbon. This view was criticized by Schmidlin,3 who contended that they should be regarded as molecular compounds analogous to those formed between National Research Fellow. (a) Schlenk and Weikel, Ber., 44, 1182 (1911); (b) Schlenk and Thal, ibid., 46, 2840 (1913); (c) Schlenk, Appenrodt, MichaeI and Thal, ibid., 47, 486 (1914). 3 Schmidlin, “Das Triphenylrnethyl,” Stuttgart, 1914, pp. 186-188. 1

2